task question
DESCRIPTION
Task Question. Is it possible to monitor news media from regions all over the world over extended periods of time, extracting low-level events from them, and piece them together to automatically track and predict conflict in all the regions of the world?. The Ares project. - PowerPoint PPT PresentationTRANSCRIPT
Task Question
• Is it possible to monitor news media from regions all over the world over extended periods of time, extracting low-level events from them, and piece them together to automatically track and predict conflict in all the regions of the world?
The Ares project
OnlineInformationSources
RiceEventData
Extractor
Singularitydetection
Hubs &Authorities
Models
AP, AFP,BBC, Reuters,…
Over 1 millionarticles on theMiddle East from1979 to 2005 (filtered automatically)
http://ares.cs.rice.edu
Analysis of wire stories
Date Actor Target Weis Code Wies event Goldstein scale790415 ARB ISR 223 (MIL ENGAGEMENT) -10790415 EGY AFD 194 (HALT NEGOTIATION) -3.8790415 PALPL ISR 223 (MIL ENGAGEMENT) -10790415 UNK ISR 223 (MIL ENGAGEMENT) -10790415 ISR EGY 31 (MEET) 1790415 EGY ISR 31 (MEET) 1790415 ISRMIL PAL 223 (MIL ENGAGEMENT) -10790415 PALPL JOR 223 (MIL ENGAGEMENT) -10790415 EGY AFD 193 (CUT AID) -5.6790415 IRQ EGY 31 (MEET) 1790415 EGY IRQ 31 (MEET) 1790415 ARB CHR 223 (MIL ENGAGEMENT) -10790415 JOR AUS 32 (VISIT) 1.9790415 UGA CHR 32 (VISIT) 1.9790415 ISRGOV ISRSET 54 (ASSURE) 2.8
Singularity detectionon aggregated eventsdata
Hubs and authoritiesanalysis of eventsdata
Relevance filter
Embedded learner design• Representation
– Identify relevant stories, extract event data from them, build time series models and graph-theoretic models.
• Learning– Identifying regime shifts in events data, tracking
evolution of militarized interstate disputes (MIDs) by hubs/authorities analysis of events data
• Decision-making– Issuing early warnings of outbreak of MIDs
Identifying relevant stories
– Only about 20% of stories contain events that are to be extracted.
• The rest are interpretations, (e.g., op-eds), or are events not about conflict (e.g., sports)
– We have trained Naïve Bayes (precision 86% and recall 81%), SVM classifiers (precision 92% and recall 89%) & Okapi classifiers (precision 93% and recall 87%) using a labeled set of 180,000 stories from Reuters.
– Surprisingly difficult problem!• Lack of large labeled data sets; • Poor transfer to other sources (AP/BBC)• The category of “event containing stories” is not well-separated
from others, and changes with time
Lee, Tran, Singer, Subramanian, 2006
Okapi classifier• Reuters data set: relevant
categories are GVIO, GDIP, G13; irrelevant categories: 1POL, 2ECO, 3SPO, ECAT, G12, G131, GDEF, GPOL
Irr
Rel New article
Decision rule: sum of top N Okapi scores in Rel set > sum of top N Okapi scores in Irr set then classify as rel; else irr
Okapi measure takestwo articles and givesthe similarity between them.
Event extraction
Parse sentence
Klein and Manning parser
Pronoun de-referencing
Sentence fragmentation
Correlative conjunctions
Extract embedded sentences (SBAR)
Conditional random fields
We extract who (actor) did what (event) to whom (target)
Not exactly the same as NER
Results
200 Reuters sentences; hand-labeled with actor, target,and event codes (22 and 02).
TABARIis stateof the artcoderin politicalscience
Stepinksi, Stoll, Subramanian 2006
Events data
177,336 events from April 1979 to October 2003 in Levantdata set (KEDS).
Date Actor Target Weis Code Wies event Goldstein scale790415 ARB ISR 223 (MIL ENGAGEMENT) -10790415 EGY AFD 194 (HALT NEGOTIATION) -3.8790415 PALPL ISR 223 (MIL ENGAGEMENT) -10790415 UNK ISR 223 (MIL ENGAGEMENT) -10790415 ISR EGY 31 (MEET) 1790415 EGY ISR 31 (MEET) 1790415 ISRMIL PAL 223 (MIL ENGAGEMENT) -10790415 PALPL JOR 223 (MIL ENGAGEMENT) -10790415 EGY AFD 193 (CUT AID) -5.6790415 IRQ EGY 31 (MEET) 1790415 EGY IRQ 31 (MEET) 1790415 ARB CHR 223 (MIL ENGAGEMENT) -10790415 JOR AUS 32 (VISIT) 1.9790415 UGA CHR 32 (VISIT) 1.9790415 ISRGOV ISRSET 54 (ASSURE) 2.8
What can be predicted?
Singularity detection
Stoll and Subramanian, 2004, 2006
Singularities = MID start/end
biweek
Date range
event17-35 11/79 to
8/80Start of Iran/Iraq war
105-111 4/83 to 7/83 Beirut suicide attack, end of Iran/Iraq war
244 1/91 to 2/91 Desert Storm
413-425 1/95 to 7/95 Rabin assassination/start of Intifada
483-518 10/97 to 2/99
US/Iraq confrontation via Richard Butler/arms inspectors
522-539 4/99 to 11/99
Second intifada Israel/Palestine
Interaction graphs
• Model interactions between countries in a directed graph.
Date Actor Target Weis Code Wies event Goldstein scale790415 ARB ISR 223 (MIL ENGAGEMENT) -10790415 EGY AFD 194 (HALT NEGOTIATION) -3.8790415 PALPL ISR 223 (MIL ENGAGEMENT) -10790415 UNK ISR 223 (MIL ENGAGEMENT) -10790415 ISR EGY 31 (MEET) 1
ARB ISR
EGY UNK
AFD PALPL
Hubs and authorities for events data
• A hub node is an important initiator of events.• An authority node is an important target of
events.• Hypothesis:
– Identifying hubs and authorities over a particular temporal chunk of events data tells us who the key actors and targets are.
– Changes in the number and size of connected components in the interaction graph signal potential outbreak of conflict.
Hubs/Authorities picture of Iran Iraq war
2 weeks prior to Desert Storm
Validation using MID data• Number of bi-weeks with MIDS in Levant data: 41 out of
589.• Result 1: Hubs and Authorities correctly identify actors
and targets in impending conflict.• Result 2: Simple regression model on change in hubs
and authorities scores, change in number of connected components, change in size of largest component 4 weeks before MID, predicts MID onset.
• Problem: false alarm rate of 16% can be reduced by adding political knowledge of conflict.
Stoll and Subramanian, 2006
Current work
• Extracting economic events along with political events to improve accuracy of prediction of both economic and political events.
Publications
• An OKAPI-based approach for article filtering, Lee, Than, Stoll, Subramanian, 2006 Rice University Technical Report.
• Hubs, authorities and networks: predicting conflict using events data, R. Stoll and D. Subramanian, International Studies Association, 2006 (invited paper).
• Events, patterns and analysis, D. Subramanian and R. Stoll, in Programming for Peace: Computer-aided methods for international conflict resolution and prevention, 2006, Springer Verlag, R. Trappl (ed).
• Four Way Street? Saudi Arabia's Behavior among the superpowers, 1966-1999, R. Stoll and D. Subramanian, James A Baker III Institute for Public Policy Series, 2004.
• Events, patterns and analysis: forecasting conflict in the 21st century, R. Stoll and D. Subramanian, Proceedings of the National Conference on Digital Government Research, 2004.
• Forecasting international conflict in the 21st century, D. Subramanian and R. Stoll, in Proc. of the Symposium on Computer-aided methods for international conflict resolution, 2002.
The research team